SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.08231 · CVPR · 2023

Rotation-Invariant Transformer for Point Cloud Matching

Zheng Qin, T Munich, Hao Yu, Mahdi Saleh, Benjamin Busam, Slobodan Ilic, Dongsheng Li, Meta Labs, Ji Hou, 3dwe Ai 5 Siemens

arXiv · PDF · Open in the Atlas

Code that ran

9 of 10 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 6 have no contract verdict; 3 failures of our instrument; 1 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2303.08231.json · badge: /api/ran/2303.08231.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
haoyu94/roitr canonical (classifier label)
classifier basis not recorded on the link
9 ran · 1 no recorded run
FunctionStatusWhere it lives
get_activation Ran
ran on a synthesized input
haoyu94/roitr/model/utils.py
code served (permissive licence) · get_code("eaf1da4c91f1180b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned LeakyReLU. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# The function get_activation takes:
# 1. activation (str): one of 'relu', 'lrelu', 'elu', 'sigmoid', 'gelu'
# 2. **kwargs: optional parameters like 'negative_slope' for 'lrelu'

# Test with 'lrelu' activation which uses kwargs
ARGS = {
    'activation': 'lrelu',
    'negative_slope': 0.02
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type LeakyReLU — a LeakyReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a LeakyReLU output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_batched_square_distance_matrix Ran
ran on a synthesized input
haoyu94/roitr/dataset/common.py
code served (permissive licence) · get_code("98abf1f4475a7452")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family float, shape (2, 4, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create small point clouds
# src: [B=2, N=4, 3] - batch of 2, with 4 source points in 3D
src = np.random.randn(2, 4, 3).astype(np.float32)

# tgt: [B=2, M=5, 3] - batch of 2, with 5 target points in 3D
tgt = np.random.randn(2, 5, 3).astype(np.float32)

ARGS = {
    "src": src,
    "tgt": tgt,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (2, 4, 5) — shape (2, 4, 5) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_dropout Ran
ran; our contract draft was wrong, not the code
haoyu94/roitr/model/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from zhiyuanYU134/MIRETR, is pointer only (licence: NONE) · get_code("8629de50bd3cc48f")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from zhiyuanYU134/MIRETR, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_square_distance_matrix Ran
ran on a synthesized input
haoyu94/roitr/dataset/common.py
code served (permissive licence) · get_code("4dcf072611e9da6a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (4, 2) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create small point clouds for source and target
np.random.seed(0)

src = np.random.randn(4, 3).astype(np.float32)  # 4 points, 3D
tgt = np.random.randn(2, 3).astype(np.float32)  # 2 points, 3D

ARGS = {
    "src": src,
    "tgt": tgt,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (4, 2) — shape (4, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_config Ran
ran; our contract draft was wrong, not the code
haoyu94/roitr/configs/utils.py
code served (permissive licence) · get_code("95ce4f03b8524dff")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tempfile
import os
import yaml

# Create a temporary YAML config file
temp_dir = tempfile.mkdtemp()
config_file = os.path.join(temp_dir, "test_config.yaml")

# Create a valid YAML config with nested dictionaries
config_content = {
    "section1": {
        "param1": "value1",
        "param2": 42
    },
    "section2": {
        "param3": 3.14,
        "param4": True
    }
}

with open(config_file, 'w') as f:
    yaml.dump(config_content, f)

ARGS = {
    "path": config_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
yaml — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_info Ran
ran on a synthesized input
haoyu94/roitr/dataset/common.py
code served (permissive licence) · get_code("5b17d30ad5d8ca5d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pickle
import tempfile
import os

# Create a temporary directory and file for the pickle
temp_dir = tempfile.mkdtemp()
pickle_path = os.path.join(temp_dir, "test_info.pkl")

# Create a sample dictionary with dataset information
sample_info = {
    "dataset_name": "test_dataset",
    "num_classes": 10,
    "num_samples": 1000,
    "feature_dim": 128,
    "split": "train",
    "metadata": {
        "version": "1.0",
        "created_by": "test"
    }
}

# Write the dictionary to a pickle file
with open(pickle_path, 'wb') as f:
    pickle.dump(sample_info, f)

# Define ARGS
ARGS = {
    "path": pickle_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
natural_key Ran
ran; our contract draft was wrong, not the code
haoyu94/roitr/lib/utils.py
code served (permissive licence) · get_code("d39e8d00511fe418")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import random
import time
import re

def natural_key(string_):
    """
    Sort strings by numbers in the name
    """
    return [int(s) if s.isdigit() else s for s in re.split('(\\d+)', string_)]

# The function takes a single parameter: string_
# It's used to split by digits and convert digit strings to integers
# We need to provide a valid string argument

ARGS = {
    "string_": "file123name456test"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
read_entries Ran
ran on a synthesized input
haoyu94/roitr/lib/utils.py
code served (permissive licence) · get_code("2db146b9e3d5e5d5")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import random
import tempfile
import os
import numpy as np

# Create a temporary directory structure with .npz files
temp_dir = tempfile.mkdtemp()
train_dir = os.path.join(temp_dir, 'train', 'subdir')
val_dir = os.path.join(temp_dir, 'val', 'subdir')
test_dir = os.path.join(temp_dir, 'test', 'subdir')

os.makedirs(train_dir, exist_ok=True)
os.makedirs(val_dir, exist_ok=True)
os.makedirs(test_dir, exist_ok=True)

# Create some dummy .npz files
np.savez(os.path.join(train_dir, 'sample1.npz'), data=np.array([1, 2, 3]))
np.savez(os.path.join(train_dir, 'sample2.npz'), data=np.array([4, 5, 6]))
np.savez(os.path.join(val_dir, 'sample1.npz'), data=np.array([7, 8, 9]))
np.savez(os.path.join(test_dir, 'sample1.npz'), data=np.array([10, 11, 12]))

ARGS = {
    'split': 'train',
    'data_root': temp_dir,
    'shuffle': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
to_tsfm Ran
ran on a synthesized input
haoyu94/roitr/lib/utils.py
code served (permissive licence) · get_code("10e533fa69e03fc9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (4, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a 3x3 rotation matrix (identity for simplicity)
rot = np.eye(3)

# Create a 3x1 translation vector
trans = np.array([[1.0], [2.0], [3.0]])

ARGS = {
    "rot": rot,
    "trans": trans
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (4, 4) — shape (4, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
weighted_circle_loss Not yet run
unverified (harvested, no recorded run)
haoyu94/roitr/lib/loss.py
code served (permissive licence) · get_code("93b0aaa0a9f9ff25")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 4 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 3 later) and 6 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
get_dropout 2 other papers
1 dated earlier, 1 later · 1 only as an assembly brick
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
2404.04557 (2024-04)
Learning Instance-Aware Correspondences for Robust Multi-Instance Point Cloud Registration in Cluttered Scenes
natural_key 2 other papers
1 dated earlier, 1 later
2203.10977 (2022-03)
Improving anatomical plausibility in medical image segmentation via hybrid graph neural networks: applications to chest x-ray analysis
2406.12837 (2024-06)
LayerMerge: Neural Network Depth Compression through Layer Pruning and Merging
load_config 1 other paper
1 dated earlier
2111.12591 (2021-11)
Lepard: Learning partial point cloud matching in rigid and deformable scenes
weighted_circle_loss 1 other paper
1 later
2407.10542 (2024-07)
3D Geometric Shape Assembly via Efficient Point Cloud Matching

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The intrinsic rotation invariance lies at the core of matching point clouds with handcrafted descriptors. However, it is widely despised by recent deep matchers that obtain the rotation invariance extrinsically via data augmentation. As the finite number of augmented rotations can never span the continuous SO(3) space, these methods usually show instability when facing rotations that are rarely seen. To this end, we introduce RoITr, a Rotation-Invariant Transformer to cope with the pose variations in the point cloud matching task. We contribute both on the local and global levels. Starting from the local level, we introduce an attention mechanism embedded with Point Pair Feature (PPF)-based coordinates to describe the poseinvariant geometry, upon which a novel attention-based encoder-decoder architecture is constructed. We further propose a global transformer with rotation-invariant crossframe spatial awareness learned by the self-attention mechanism, which significantly improves the feature distinctiveness and makes the model robust with respect to the low overlap. Experiments are conducted on both the rigid and non-rigid public benchmarks, where RoITr outperforms all the state-of-the-art models by a considerable margin in the low-overlapping scenarios. Especially when the rotations are enlarged on the challenging 3DLoMatch benchmark, RoITr surpasses the existing methods by at least 13 and 5 percentage points in terms of Inlier Ratio and Registration Recall, respectively. Code is publicly available 1 .

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2303.08231")
get_code_for_paper("2303.08231")
have("2303.08231")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2303.08231.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2303.08231.svg)](https://syntology.ai/paper/2303.08231)

Connect an agent — have() is free.